Re: [v6ops] RFC 2119 language in an informational draft?

["STARK, BARBARA H" <[e-mail]>]

> > >> May I suggest that an informational draft doesn't impose requirements,
> > so there is no need for RFC 2119 language?
> > >> I think that that suggestion is erroneous.
> > >
> > > +1.
> > >
> > > While this might be argumentum ad antiquitam, it is clear that this
> > isn't the standard interpretation.
> > >
> > > Looking through the RFCs 8000 and above, I find 262 Informational
> > documents using RFC2119 (RFC8174) language, and >4000 instances:
> > > $ for file in (grep -l 'Category: Informational' rfc8???.txt)
> > >                                                   egrep
> > "SHOULD|MUST|REQUIRED|SHALL|RECOMMENDED|MAY|OPTIONAL" $file
> > >                                               end >
> > > ~/tmp/info_2119.txt $ wc -l ~/tmp/info_2119.txt
> > >   4023
> >
> > [pedantic mode on] An informational document may contain quotes from the
> > standard ones, so your grep would count them as well.
> >
> Hello Jen,
> 
> In that case, maybe is makes sense to allow the MUST (should it be between
> quotes?) but not to allow the BCP 14 snippet.
> 
> Now, can there be BCP 14 text in an informational that would go straight
> through AD sponsorship, and document a protocol that was defined outside the
> IETF, e.g., implemented by one company and offered to the community? I guess
> uppercase makes sense too in such document, and the BCP 14 snippet does too.
> 
> No later than yesterday I received a proposed amendment, for an informational
> text, that contained MUST and SHOULD. So yes, the mistake must be
> commonplace as people are not necessarily aware of the practice.
> 
> Net-Net: I see the value of raising a warning in a nit check, and document the
> exception in the shepherd write up if the uppercase is confirmed, but not that
> we can bar it the hard way.

I'm not aware of any prohibition on having normative language in an informational draft and would be very much opposed to such a prohibition. For example, RFCs 7084 (IPv6 CE router), 8585 (IPv6 transition CE router), and 2516 (PPPoE) would be seriously diminished by a lack of RFC2119 language. The first 2 of these are v6ops products.
Barbara